Skip to main content

What is the Biolevate API?

Getting Started

This guide introduces the Biolevate API and its core concepts. For hands-on examples, see the Getting Started guide.

The Elise Platform​

Elise is Biolevate's document intelligence platform. It connects to your storage backends, indexes files with AI-powered analysis, and extracts structured metadata (ontologies) from your documents. The Biolevate API provides programmatic access to this platform through a REST interface.

How It Works​

The API is organized around several resource types that form a natural pipeline:

Providers​

Providers represent external storage backends connected to the Elise platform. They can be Amazon S3 buckets, Azure Blob Storage containers, Google Cloud Storage buckets, SharePoint Online document libraries, or local file systems.

Providers are configured through the Elise admin portal. The API exposes read-only access to list and inspect them.

Provider Items​

Provider Items are the raw files and folders that exist inside a Provider. Through the API you can:

  • Browse directories and list files
  • Upload files (direct or via presigned URLs for large files)
  • Create folders
  • Rename and delete items
  • Download files

Operating on provider items does not trigger any indexation or analysis. They are raw storage operations.

EliseFile​

An EliseFile is the central resource. When you create an EliseFile from a provider item, the platform:

  1. Registers the file in the Elise system
  2. Triggers indexation to parse and analyze the document content
  3. Computes ontologies (AI-extracted metadata and concepts)

This is the step that transforms a raw file into an intelligent, searchable document.

Collections​

Collections are logical groups of EliseFiles. They allow you to organize indexed documents into workspaces, projects, or any grouping that fits your workflow. A file can belong to multiple collections.

Question Answering​

The Question Answering API lets you submit natural-language questions about your indexed documents. The platform processes them as asynchronous jobs and returns AI-generated answers with explanations and source references pointing back to the document content.

Extraction​

The Extraction API lets you define named metadata fields and have the AI extract their typed values from your indexed documents. Like Question Answering, it runs as an asynchronous job and returns structured results with source references.

Multi-Dimensional Extraction​

The Multi-Dimensional Extraction API lets you extract structured entity tables from indexed documents. Instead of asking for independent metadata fields, you define a schema of related columns, such as identifier, group, timepoint, value, unit, or comment, and the API returns rows where those values belong together.

Use it when the answer is naturally tabular: clinical outcomes by treatment arm and week, compounds with dose and unit, endpoints with definitions, or measurements split across groups and timepoints.

Find Similar Files​

The Find Similar Files API lets you submit source identifiers, such as DOIs or open access IDs, and search for matching files locally and through remote bibliographic sources such as Arxiv and PubMed. If a remote record is found but no open access file is available, the API returns a metadata-only match.

Agent​

The Agent API runs an iterative reasoning loop over your indexed documents and conversation history. Instead of a single question or a fixed set of fields, you send a message (or a list of messages) and the agent decides its own steps to produce an answer. It runs as an asynchronous job and can return either plain text or a structured answer constrained by a JSON Schema. It supports stateless runs (full message list) and stateful conversations (a new message plus a conversationId).

Supported File Formats​

The API supports indexation for the following file types:

CategoryFormats
DocumentsPDF, DOC, DOCX, TXT, HTML, HTM
SpreadsheetsXLSX, XLS, CSV, ODS
ImagesJPG, JPEG, PNG, WEBP, BMP, TIFF
DataJSON, XML

The REST API​

All endpoints are served under the /api/core/ prefix and require Bearer token authentication via a Personal Access Token (PAT).

ResourceOperationsDescription
ProvidersList, GetBrowse connected storage backends
Provider ItemsList, Upload, Create, Rename, Delete, DownloadManage files and folders in storage
FilesList, Create, Get, Delete, Reindex, OntologiesIndex documents and extract metadata
Question AnsweringCreate job, Poll, Results, Inputs, Annotations, ListAI-generated answers to questions about document content
ExtractionCreate job, Poll, Results, Inputs, Annotations, ListAI-extracted structured metadata fields from documents
Multi-Dimensional ExtractionCreate job, Poll, Results, Inputs, Annotations, ListAI-extracted entity rows and cells from documents
Find Similar FilesCreate job, Poll, ListLocal and remote matching from source identifiers
AgentCreate job, Poll, Results, Inputs, Annotations, ListConversational reasoning over documents and conversation history
AnnotationsGet (per job)Source passages and positions referenced by QA, Extraction, Multi-Dimensional Extraction, and Agent results
CollectionsCRUD + file associationOrganize indexed documents into groups
API Reference

For complete endpoint documentation with request/response schemas, see the API Reference.

Next Steps​

  1. Get authenticated and make your first API call
  2. Browse your providers to discover connected storage
  3. Manage files in your storage backends
  4. Index documents to unlock AI-powered metadata
  5. Organize into collections for structured workflows
  6. Ask questions about document content with AI
  7. Extract metadata fields from documents with AI
  8. Extract entity rows using a schema
  9. Find similar files from source identifiers
  10. Run an agent over your documents and conversation history